An answer recommendation method and device

Abstract

The present invention provides an answer recommendation method and device, wherein, the method includes: obtaining the text content of the question and the answer corresponding to the question, segmenting words to obtain the semantic unit of the question and the semantic unit of the answer; using the pre-established question domain dictionary, find out the weights of the semantic units of the questions in each category, and calculate the topic weights of the questions in each category; and, use the pre-established answer domain dictionary to find out the semantic units of the answers in weights in each category, respectively calculate the topic weights of the answers in each category; use the topic weights of the obtained questions and the topic weights of each answer to calculate the topic similarity between each answer and the question, according to The calculation result of the topic similarity recommends an answer. Compared with the prior art, the present invention generates a question domain dictionary and an answer domain dictionary respectively, which effectively improves the accuracy rate of question and answer for semantic similarity and improves the recall rate.

Description

【Technical field】 [0001] The invention relates to the technical field of Internet information processing, in particular to an answer recommendation method and device. 【Background technique】 [0002] With the continuous development of information and network technology, online interactive question-and-answer communities such as Baidu Zhizhi, Sina Aiwen, Google Q&A, Soso Ask, Yahoo Knowledge Hall, etc., are increasingly attracting people's attention. These online interactive question-and-answer communities provide a platform for netizens to interact and communicate. Users can freely ask questions, browse questions, answer questions, communicate with each other, and share knowledge. With the increasing number of users participating in the question-and-answer community, the number of candidate answers increases accordingly, and the question-and-answer community usually automatically sorts the answers in order to recommend optimal answers for users. [0003] In the automatic sor...

AI Technical Summary

Problems solved by technology

However, questions and answers in question-and-answer pairs may be described in different ways, that is, there may be inconsistencies in the use of words. system, etc.; while the domain word distribution of the answer is mainly some professional computer vocabulary, such as PC, win7, etc.; another example, the user asks a question about the skills of a certain game, but the answer answered by the user is specific to the skill. description, not including the words used in the question

At this time, the semantic correlation between the answer and the question calculated according to the existing method is low, which will make it impossible to recall the answer that actually matches the question or the ranking of the answer is low, resulting in a decline in the accuracy of the quality judgment of the question and answer, making users Could not find a preferred answer

Images